[CIR][ABI][AArch64][Lowering] Support unions (#1075)

As the title says, this PR adds support for unions for AArch64 lowering.
The idea is basically the same as the
[original](https://github.com/llvm/clangir/blob/dbf320e5c3db0410566ae561067c595308870bad/clang/lib/AST/RecordLayoutBuilder.cpp#L2111)
codegen, and I added a couple of tests.

Author: bruteforceboy

clang/lib/CIR/Dialect/Transforms/TargetLowering/RecordLayoutBuilder.cpp

@@ -437,9 +437,9 @@ void ItaniumRecordLayoutBuilder::layoutField(const Type D,
 
   // Reserve space for this field.
   if (!IsOverlappingEmptyField) {
-    // uint64_t EffectiveFieldSizeInBits = Context.toBits(EffectiveFieldSize);
+    uint64_t EffectiveFieldSizeInBits = Context.toBits(EffectiveFieldSize);
     if (IsUnion)
-      cir_cconv_unreachable("NYI");
+      setDataSize(std::max(getDataSizeInBits(), EffectiveFieldSizeInBits));
     else
       setDataSize(FieldOffset + EffectiveFieldSize);
 

clang/test/CIR/CallConvLowering/AArch64/union.c

@@ -0,0 +1,41 @@
+// RUN: %clang_cc1 -triple aarch64-unknown-linux-gnu  -fclangir -fclangir-call-conv-lowering -emit-cir-flat -mmlir --mlir-print-ir-after=cir-call-conv-lowering %s -o %t.cir
+// RUN: FileCheck --input-file=%t.cir %s -check-prefix=CIR
+// RUN: %clang_cc1 -triple aarch64-unknown-linux-gnu  -fclangir -emit-llvm %s -o %t.ll -fclangir-call-conv-lowering
+// RUN: FileCheck --input-file=%t.ll %s -check-prefix=LLVM
+
+// CIR: !ty_U = !cir.struct<union "U" {!s32i, !s32i, !s32i}>
+// LLVM: %union.U = type { i32 }
+typedef union {
+  int a, b, c;
+} U;
+
+// CIR: cir.func @foo(%arg0: !u64i
+// CIR: %[[#V0:]] = cir.alloca !ty_U, !cir.ptr<!ty_U>, [""] {alignment = 4 : i64}
+// CIR: %[[#V1:]] = cir.cast(integral, %arg0 : !u64i), !u32i
+// CIR: %[[#V2:]] = cir.cast(bitcast, %[[#V0]] : !cir.ptr<!ty_U>), !cir.ptr<!u32i>
+// CIR: cir.store %[[#V1]], %[[#V2]] : !u32i, !cir.ptr<!u32i>
+// CIR: cir.return
+
+// LLVM: void @foo(i64 %[[#V0:]]
+// LLVM: %[[#V2:]] = alloca %union.U, i64 1, align 4
+// LLVM: %[[#V3:]] = trunc i64 %[[#V0]] to i32
+// LLVM: store i32 %[[#V3]], ptr %[[#V2]], align 4
+// LLVM: ret void
+void foo(U u) {}
+
+// CIR: cir.func no_proto @init() -> !u32i
+// CIR: %[[#V0:]] = cir.alloca !ty_U, !cir.ptr<!ty_U>, ["__retval"] {alignment = 4 : i64}
+// CIR: %[[#V1:]] = cir.load %[[#V0]] : !cir.ptr<!ty_U>, !ty_U
+// CIR: %[[#V2:]] = cir.cast(bitcast, %[[#V0]] : !cir.ptr<!ty_U>), !cir.ptr<!u32i>
+// CIR: %[[#V3:]] = cir.load %[[#V2]] : !cir.ptr<!u32i>, !u32i
+// CIR: cir.return %[[#V3]] : !u32i
+
+// LLVM: i32 @init()
+// LLVM: %[[#V1:]] = alloca %union.U, i64 1, align 4
+// LLVM: %[[#V2:]] = load %union.U, ptr %[[#V1]], align 4
+// LLVM: %[[#V3:]] = load i32, ptr %[[#V1]], align 4
+// LLVM: ret i32 %[[#V3]]
+U init() {
+  U u;
+  return u;
+}